Hi, On Mon, Jul 12, 2010 at 07:07:19PM +0200, Samuel Ortiz wrote: > On Fri, Jul 09, 2010 at 02:39:05PM -0400, Forest Bond wrote: > > Hi Samuel, > > > > On Thu, Jul 08, 2010 at 07:35:38PM +0200, Samuel Ortiz wrote: > > > So, I am attaching 2 patches to this email: > > > > > > 0001-Factorize-host-route-setting-routine.patch that cleans the inet.c > > > host > > > route setting stuff and that should go upstream anyway. > > > > > > And > > > 0002-Set-DNS-host-routes-before-toggling-the-service-READ.patch that > > > basically > > > sets the DNS host routes before we set the service to READY, from > > > connection_probe(). Once the routes are properly set, bumping the service > > > to > > > READY will trigger an update_nameservers() call, which should allow our > > > DNS > > > proxy to succesfully connect to the nameservers. > > > > > > Would you mind giving those 2 patches a try, please ? They should apply > > > cleanly on top of the latest ConnMan git. > > > > These work fine, as far as I can tell. > > > > One question: should the route always be added with a gateway? If the > > nameserver is on the same subnet as the connected interface, doesn't that > > cause > > DNS requests to be routed through the gateway unnecessarily? > Well, if the nameserver is on the same subnet as the connected interface then > we actually won't need host routes to it. When setting the IP address, the > subnet route is set and from there you'd have a route to your namserver.
Right, but your patch would add host routes in that case anyway, right? I'm just wondering if doing that affects how packets are routed to those nameservers. Thanks, Forest -- Forest Bond http://www.alittletooquiet.net http://www.pytagsfs.org
signature.asc
Description: Digital signature
_______________________________________________ connman mailing list [email protected] http://lists.connman.net/listinfo/connman
